The Walled Web Newsletter

Fact check: are these five companies blocking AI?

In October 2026 a post by @trikcode named five companies moving against AI agents, the AI tools that browse, shop and book for you. @gregisenberg shared it with the line "The war on agents has begun". Here is what each company has actually said, and what its public files show today.

Every source below was opened and checked against what this page says on 19 Oct 2026. Companies change their minds. If something here is out of date or wrong, tell us and we will publish the correction.

Scorecard, 19 Oct 2026

CompanyWhat it saidrobots.txtFront doorOur result
Delta Air LinesNo deal for outside booking agentsNo AI rulesLet our bot inBlocks some
JPMorgan ChaseBlocks agents using customer loginsNo AI rulesLet our bot inBlocks some
Charles SchwabBlocks unauthorized outside agentsCould not read itTurned our bot awayBlocks AI
YelpAgents must pay to licenseBlocks 24 of 24 AI botsDid not try
AmazonUnauthorized agents violate its termsBlocks 17 of 24 AI botsLet our bot inBlocks some

robots.txt counts are out of the 24 AI bots we track. Front door: what happened when our bot, which says who it is, visited the homepage. Details and sources for each company are below.

Delta Air Lines

"Delta is blocking AI agents from booking flights."

Partly supported
What the company said
A Delta spokesperson told TechCrunch that Delta "does not currently have a partnership or integration that enables a third-party AI agent to shop for or book Delta flights," is "continuing to evaluate" outside agents, and takes steps against unauthorized automated activity. That is a statement about partnerships and bot defense, not an announced block.
What its public files show
Blocks some as of 19 Oct 2026, 13:45 UTC

Its robots.txt (the public rules file for bots) has no rules for any AI bot we track. They are covered by its catch-all rule (User-agent: *), which lets them into the site. But Delta Air Lines says it has no partnership or integration that lets a third-party AI agent book its flights, and that it takes steps against unauthorized automated activity (TechCrunch, 6 Oct 2026).

Front door: Let in HTTP 200.

What this can't show
Booking happens behind search forms and checks that our bot never touches. A block on booking agents would not show up in robots.txt.
JPMorgan Chase

"JPMorgan Chase is blocking agents from accessing accounts."

Supported by the company
What the company said
On the bank's third-quarter earnings call, Jamie Dimon said: "We're not going to allow just any outside agent to come inside the company." A spokesperson said the bank blocks outside AI agents that log in with customers' banking credentials, and that it supports "secure, customer-authorized data sharing through open banking providers." CNBC reports that this lets agents read some account data, such as balances, but not move money.
What its public files show
Blocks some as of 19 Oct 2026, 13:29 UTC

Its robots.txt (the public rules file for bots) has no rules for any AI bot we track. They are covered by its catch-all rule (User-agent: *), which lets them into the site. But JPMorgan Chase says it blocks outside AI agents that log in with customers' banking credentials (CNBC, 13 Oct 2026).

Front door: Let in HTTP 200.

What this can't show
Blocks on account access happen after login, where no public file can see them. We never log in, so we can confirm what the company said, not how the block works.
Charles Schwab

"Charles Schwab is blocking agents from accessing accounts."

Supported by the company
What the company said
Schwab blocks unauthorized outside AI agents that use client login credentials or browser automation, and CEO Rick Wurster said outside AI will not be allowed to move money. Schwab says it will let outside AI tools connect through Schwab-controlled MCP servers to check balances and do research, and it is steering clients to its own assistant, Charley.
What its public files show
Blocks AI as of 19 Oct 2026, 13:46 UTC

We could not read this site's robots.txt, its public rules file for bots (a refusal page sent with HTTP 200 instead of an error code). A refusal like this usually comes from anti-bot software, not from the site's written rules, so on its own it does not count as a block. The homepage also turned our bot, which says who it is, away with a refusal page sent with HTTP 200 instead of an error code. Charles Schwab says it blocks unauthorized outside AI agents that log in with client credentials or drive a browser (American Banker, 15 Oct 2026). Under our rules, that statement plus our bot being turned away sets the status to Blocks AI.

Front door: Turned away HTTP 200, a refusal page sent with HTTP 200 instead of an error code.

What this can't show
Blocks on account access happen after login, where no public file can see them. We never log in, so we can confirm what the company said, not how the block works.
Yelp

"Yelp requires paid data licensing for agents."

Supported by the company
What the company said
Yelp told TechCrunch that it doesn't permit non-human traffic on its platform unless the agent has paid to access Yelp's content and data through its data licensing program. Its robots.txt carries a section headed "AI / LLM Crawlers and Agents."
What its public files show
as of 19 Oct 2026, 13:31 UTC

Yelp says AI agents may use the site only through its paid data licensing program. Its robots.txt (the public rules file for bots) tells all 24 AI bots we track to stay out of the whole site. It names GPTBot, ClaudeBot, Google-Extended, CCBot, Meta-ExternalAgent and 6 more directly. A catch-all rule (User-agent: *) keeps the rest out too. The same rules ask our bot to stay out, so we read only robots.txt.

What this can't show
Licensing deals are private. We can show the public rules and what the company said, not who has paid.
Amazon

"Amazon is blocking Grok and other shopping agents."

Partly supported
What the company said
For other shopping agents, there is a public record: in September 2026 Amazon showed Meta's Muse agent a notice that "continued access by an unauthorized AI agent violates Amazon's Conditions of Use", and it is in litigation with Perplexity over the Comet browser agent. For Grok specifically, the only evidence found is an X post; Amazon did not respond to a request for comment. Amazon's robots.txt does disallow Grok-style tokens (GrokBot, xAI-Grok, Grok-DeepSearch), but we found no xAI documentation of any such crawler.
What its public files show
Blocks some as of 19 Oct 2026, 13:23 UTC

Its robots.txt (the public rules file for bots) treats AI bots differently from each other. It blocks 17 of the 24 AI bots we track from the whole site, including GPTBot, ClaudeBot, Google-Extended, CCBot and Meta-ExternalAgent. Applebot-Extended can reach most of it. Amazon told Meta's shopping agent that access by an unauthorized AI agent violates its Conditions of Use (The Next Web, 21 Sep 2026).

Its robots.txt names Grok-DeepSearch, GrokBot, xAI-Grok. xAI, the company behind Grok, has never published a bot by these names.

Front door: Let in HTTP 200.

What this can't show
Whether Amazon stops Grok at checkout cannot be seen from public files; that needs reports from people who tried.

What we found

For all five, the blocking the post describes happens outside robots.txt: at login, in the terms of service, through a paid license, or in court. The banks say they block AI tools at login. Amazon told Meta's AI agent it was breaking Amazon's Conditions of Use. Yelp says AI companies must pay through a licensing program. Public files can show only part of this. That is why we also visit each site with our own bot, and why we ask people to report what their AI tools run into.